A captioning team is choosing between two Azure AI Speech features. How does speech-to-text differ from speech translation?
Show answer and explanation
Answer: A
Both features start from audio and both rely on the same recognition engine, but they stop at different points. Speech-to-text returns a transcript in the language that was actually spoken. Speech translation takes the recognised words and adds a machine translation step, so the caller receives text or synthesised speech in a chosen target language. A captioning tool for one language only needs transcription, while a multilingual meeting tool needs the translation step.
Why the other options are wrong
- B. Both features work with live microphone input and with stored audio files, so restricting one to recordings and the other to streaming is not a real distinction.
- C. The output formats are the reverse of this claim: plain speech-to-text returns text, and speech translation can return either translated text or synthesised translated speech, not audio alone.
- D. The outputs are materially different because translation changes the language of the result, and the features are exposed as separate operations with different request and response shapes.
Exam tip. Transcription keeps the source language; translation changes it.